THIS IS A TEST INSTANCE. ALL YOUR CHANGES WILL BE LOST!!!!
...
Here’s a list of proposed metrics for this new protocol:
NAME | TAGS | TYPE | NOTE |
CurrentLeader | type=raft-manager | dynamic gauge | -1 means UNKNOWN |
|---|---|---|---|
CurrentEpoch | type=raft-manager | dynamic gauge | 0 means UNKNOWN |
CurrentVote | type=raft-manager | dynamic gauge | -1 means not voted for anyone |
LogEndOffset | type=raft-manager | dynamic gauge | |
LogEndEpoch | type=raft-manager | dynamic gauge | |
BootTimestamp | type=raft-manager | dynamic gauge | |
State | type=raft-manager | dynamic enum | possible values: "leader", "follower", "candidate", "observer" |
NumQuorumVoters | type=raft-manager | dynamic gauge | number of cached voter connections; would never be larger than quorum-size |
ElectionLatencyMax/Avg | type=raft-manager | dynamic gauge | measured on each voter, start when becoming a candidate and end on learned or become the new leader |
ReplicationLatencyMax/Avg | type=raft-manager | dynamic gauge | measured on leader, start when appending the record and end on hwm advanced beyond |
InboundRequestPerSec | type=raft-manager, source-broker-id=[broker-id] | windowed rate | one per source |
OutboundRequestPerSec | type=raft-manager, destination-broker-id=[broker-id] | windowed rate | one per destination |
InboundChannelSize | type=raft-manager | windowed average | |
OutboundChannelSize | type=raft-manager | windowed average | |
FetchRecordsPerSec | type=raft-manager | windowed rate | apply to follower and observer only |
AppendRecordsPerSec | type=raft-manager | windowed rate | apply to leader only |
ErrorResponsePerSec | type=raft-manager, destination-broker-id=[broker-id] | windowed rate | one per destination |
TotalTimeMs | type=raft-manager, request=[request-type] | windowed average | one per inbound request type |
InboundQueueTimeMs | type=raft-manager, request=[request-type] | windowed average | one per inbound request type |
HandleTimeMs | type=raft-manager, request=[request-type] | windowed average | one per inbound request type |
OutboundQueueTimeMs | type=raft-manager, request=[request-type] | windowed average | one per inbound request type |
AvgIdlePercent | type=raft-manager | windowed average |
Client Interactions
Since this specific quorum implementation is only to be used by Kafka internally, we do not need to add new public protocols for clients. More specifically, the leader of the quorum would act as the controller of the cluster, and any client requests that requires updating the metadata (previously stored in ZK) would be interpreted as appending new record(s) to the quorum's internal log.
...